The Taxonomic Status of Genera within the Fabeae (Vicieae), with a Special Focus on Pisum

Abstract

The taxonomy of the tribe Fabeae (Vicieae) has long been problematic, but an analysis by Schaefer et al. in 2012 gave an exceptionally clear view of the tribe and noted the possibility that some nomenclatural adjustments may be required at some future date. These authors suggested several options, expressing some preferences. However, there has been a recent change to formally accepted names, implementing one of these possibilities, but without any additional relevant information. This change seems unjustified and unhelpful. We therefore present an argument for the retention, or re-instatement, of the genera Pisum, Vavilovia, and Lens until such time as new data support this requirement and there is no nomenclatural solution that is both accurate and convenient.

1. Introduction

Legumes (Leguminosae or Fabaceae) are a large and diverse family of angiosperms [1,2] recently expanded and reorganized to include six subfamilies [2]. Most agriculturally important legumes are found within the Papilionoideae subfamily, and within this subfamily, an informally named monophyletic group of genera, the Inverted Repeat Lacking Clade (IRLC) [3,4], includes the majority of temperate legume crop and pasture species.

The IRLC, which includes, among others, the tribes Cicereae, Wisterieae, Trifolieae, and Fabeae (Vicieae), is defined as papilionoid legume species lacking one of the two inverted repeats of the plastid genome commonly found in land plants and freshwater green algae (the Viridiplantae). As is the case for the other group of land plants where this type of plastid genome is found (twice in the Pinophyta), this structural rearrangement affects the rate of evolution of plastid genes usually located within the inverted repeat [5]. Analysis of the phylogenetics of Medicago (Trifolieae) presented evidence for a gain of an IR structure in M. minima [6]. Taken together, these observations suggest that phylogenies within the IRLC that depend on plastid genome divergence should be treated with caution.

Fabeae (Vicieae) sensu lato includes the genera Vicia, Lens, Lathyrus, Pisum, and Vavilovia [7], and a comprehensive phylogenetic analysis of this group was presented by Schaefer et al. [8], who concluded that these genera were not monophyletic, noting that Pisum and Vavilovia were nested within Lathyrus, and that Lens was nested within Vicia and furthermore, that a group of Vicia species, belonging to sections Ervum and Ervilia, were early-diverging lineages of the entire tribe. Schaefer et al. [8] discussed the taxonomic implications of this phylogeny and suggested two broad possibilities; the first was to generate a multiplicity of new genera, and the second was to amalgamate all into a single genus, Vicia. One alternative was to merge Pisum and Vavilovia with Lathyrus, and Lens with Vicia, creating three genera in the tribe, the basal genus comprised of Vicia sections Ervum and Ervilia, Vicia (excluding sections Ervum and Ervilia but including Lens), and Lathyrus, broadened to include Pisum and Vavilovia.

This three-genus scheme has been implemented, in part [9]; here, we discuss the basis and implications for this taxonomic restructuring and argue for delaying any change in the taxonomy of this tribe until more robust information is available.

2. The Fabeae

The general features of the phylogenetic tree presented by Schaefer et al. [8] are summarized in Figure 1; the data and interpretations of Schaefer et al. [8] remain the best available description of relationships within the Fabeae.

2.1. The Consensus Maximum Likelihood (ML) Phylogeny

The basal bifurcation in the consensus ML phylogeny separates Vicia-6 (V. cypria, V. lunata, V. sylvatica, V. ervilia, V. articulata, V. monantha, and V. hirsuta) from all other species, with a high degree of certainty (93% bootstrap support). It is this bifurcation and the paraphyly of Vicia that argues for the entire tribe being considered to be the monophyletic genus, Vicia. While this interpretation is well supported, it adds no information regarding the internal relationships of the Fabeae as a whole, which is provided by the recognition of additional genera within the Fabeae.

The reinstatement of the genus Ervilia, corresponding to this group of species, branching at node “i“ (Figure 1), has the potential to enable the fragmentation of the tribe into more than one genus. These species have not been renamed; for example, the name Vicia ervilia (L.) Willd. is accepted in The World Checklist of Vascular Plants (WCVP) [9]. In addition, the remaining Vicia species are paraphyletic, and the other genera are nested within Vicia.

The creation of another genus Ervum, branching at node “e” (Figure 1), has also been proposed [8]. This genus would comprise Vicia pubescens, V. tenuissima, and V. tetrasperma (Vicia-1 in Figure 1). However, this retains a paraphyletic Vicia, including Lens. The genus Ervum is not recognized by WCVP [9], but Lens culinaris is no longer accepted and has been replaced by Vicia lens (L.) Coss. & Germ. Strangely, Lens culinaris subsp. odemensis (Ladiz.) and Lens culinaris subsp. tomentosus (Ladiz.) remain [10].

The Ervum proposal creates a monophyletic group at node “d” (Figure 1) in the consensus ML phylogeny, and this group includes Pisum and Vavilovia nested within three Lathyrus lineages. However, this group has only 53% bootstrap support. Pisum and Vavilovia are distinct, each with 100% bootstrap support, even though the monospecific genus Vavilovia formosa was formerly Pisum formosum [9]. The combined Pisum plus Vavilovia is monophyletic, with 99% bootstrap support, and is thus distinct from (or within) Lathyrus.

An important issue to resolve is the status of the nodes labeled “b”, “c” and “d” (Figure 1) in the ML phylogeny. These do not have strong bootstrap support, raising the possibility that the three groups of Lathyrus species may be a monophyletic clade, rather than a paraphyletic grade, as in this figure. Supposing that Lathyrus and Pisum plus Vavilovia are two monophyletic clades derived at node “d” (Figure 1, condensing the nodes “b” and “c”, with less than 50% support, and “d”, which musters 53% support), then, in that case, there is no requirement to re-assign any generic status. On the other hand, if we accept that there are two monophyletic lineages, designated Lathyrus-1 and Lathyrus-1&2 in the chronogram of Figure 1B, then it would be possible either to elevate the smaller subgroup of Lathyrus to a new genus, or to combine Lathyrus with Pisum plus Vavilovia. The former option has four genera, and the latter has only one. Furthermore, it should be noted that the diversity within Pisum and Vavilovia is not fully represented in this tree, both with respect to the number of Pisum subspecies represented [11,12,13] and with respect to the molecular data available. Indeed, whereas Lathyrus accessions were described both by ITS and plastid DNA sequences, most Pisum accessions where only characterized by ITS sequences. A total of 15 out of 18 Pisum accessions had either one or no chloroplast sequence, which may have significant effects on the robustness of the phylogenetic relationships inferred, especially if the sampling of chloroplast data for Lathyrus is higher.

The majority of Lathyrus species are represented by Lathyrus-1 in Figure 1 and form a monophyletic group, but with 61% bootstrap support; in some trees, some of these species are placed differently, presumably in association with L. neurolobus. The third group of Lathyrus (L. articulatus, L. clymenum, and L. ochrus) is itself distinct and monophyletic, with 100% bootstrap support in the ML tree. The arrangement of these three Lathyrus groups and the Pisum plus Vavilovia group is uncertain, with the highest bootstrap support of 53% for node “d” and less than 50% support for nodes “b” and “c”, suggesting that these dichotomies may collapse when adding more data and/or taxa for further phylogenetic inferences (Figure 1).

2.2. The Chronogram

There are many useful analyses in the work of Schaefer et al. [8], one of which is a chronogram that attempted to date divergences within the Fabeae. The overall branching structure that relates to the present discussion is given in Figure 1B. There are several striking features of this structure. The first is that the proposed Ervilia group is basally branched, but the age estimate of this divergence overlaps slightly with the divergence of Vicia plus Lens from Lathyrus, Pisum, and Vavilovia. All Vicia, including the proposed Ervum, are a monophyletic lineage, but are distinct from Lens in this analysis [8]. This is consistent with retaining Lens as a separate genus.

In the chronogram, the two minor Lathyrus lineages are distinct from the remaining (and the majority of) Lathyrus species, but the age estimate for this divergence overlaps substantially with the estimated age of the divergence of Pisum plus Vavilovia from all Lathyrus. The estimate of the Pisum/Vavilovia divergence time overlaps, to some degree, with their divergence from the majority of Lathyrus, presumably reflected in the 99% bootstrap support for this clade in the ML tree.

3. Discussion

The phylogenetic analysis of the Fabeae presented by Schaefer et al. [8] is a thorough analysis of this tribe, and remains the best information we have regarding species and genus relationships in the Fabeae. These authors suggested several possible nomenclatural changes within the tribe that might better reflect the new understanding of species relationships which they uncovered, but did not present a firm and comprehensive set of new species and genus names. No new data have yet become available to lend weight to any of the possible new systems of species and genus names, yet a new system has been partially implemented by WCVP [9] and other authors [14,15,16]. This seems premature and precipitate.

Kenicer and Parsons [10] argue, “A major change is that Pisum and Vavilovia are now included in Lathyrus. If an alternative approach had been taken with Pisum and Vavilovia, maintaining them as separate genera, then Lathyrus would have to lose L. clymenum, L. neurolobus and L. ochrus. On balance the inclusion of Pisum and Vavilovia in Lathyrus entails fewer name changes and more accurately reflects the morphology of the group, meaning the inwardly hairy style is a defining character of the genus”. Whether this, in fact, entails fewer name changes depends on the number of species within Pisum plus Vavilovia. In fact, this change demands at least four name changes (Pisum sativum, P. elatius, and P. fulvum, plus Vavilovia formosa) compared to renaming four Lathyrus species. This numerical point of view is not the only consideration; a search in Web of Science [17] with the term “Pisum sativum” returned 41,916 results for this species, while a search with the term “Lathyrus” returned 4834 results for this genus, reflecting the focus of interest in one of the species to be renamed.

We have focused attention on the work of Schaefer et al. [8] because this is the most complete study in terms of species sampling. Nevertheless, additional studies have been published which are relevant to the issue of the taxonomic position of Pisum with respect to Lathyrus. Robledillo et al. [18] present a phylogeny, based on centromeric repeated sequences (which are unusual within Pisum and Lathyrus), which positions Pisum as basal to a monophyletic group which includes Lathyrus, but distinct from a monophyletic Vicia. Zhao et al. [19], based on an analysis of 15,000 nuclear genes in the Fabaceae, shows the same relationship. However, two plastid gene-based phylogenies [20,21] place Pisum within a monophyletic Lathyrus clade. In these analyses, the position of L. clymenum (and L. ochrus in [20]) is critical to the interpretation of the relationship of Pisum (and Vavilovia) to the remainder of Lathyrus, but the support for a particular branching structure is low. The branching pattern of these phylogenies may depend on the species included or excluded from the analysis and also on whether the plastid or nuclear genome is being analyzed. Similarly, the nature of the nuclear DNA sequences may again have an impact, reflecting patterns of gene duplication [19] or perhaps their function [18].

We recognize that taxonomic names should reflect our understanding of species relationships; nomenclature should above all be useful for distinguishing Organismal Taxonomical Units under study. When there are substantial changes to our understanding, a new nomenclatural system is appropriate, and this should recognize due priority, as well the value of nomenclatural stability. Pisum has been used as a name since antiquity. However, we see no good evidence that Pisum and Vavilovia (and Lens) need to be renamed. This may eventually become desirable, given new knowledge which is not yet available. If or when this situation changes, we think it would be wise to implement changes that will be minimally disruptive to the scientific literature. If the creation of a small number of new genera enables the preservation of Vicia, Lens, Lathyrus, Pisum, and Vavilovia, then that would seem a sensible change, as indeed Schaefer et al. [8] proposed for Ervum and Ervilia. Similarly, Kenicer and Parsons [14] presented a diagram of the relationships within the Fabeae, with eight taxonomic groups combined in a way that creates four genera.

If future phylogenetic analysis does confirm that there is a monophyletic clade in which Pisum and Vavilovia are indeed embedded within Lathyrus, then the creation of a new genus from within Lathyrus might enable the retention of Pisum and Vavilovia; if so, this would seem a desirable change.